Research On Detection Technology Of Image Processing Under The Hat

Currently, researching of image similarity matching based on image recognition has evolved and achieved parts of the research results. Nevertheless, in the field of online shopping searching, using the keywords as the primary key and pictures as associate key to search is still dominated, however,it has been apparently unable to meet the customers’ demand during this multimedia time. Therefore, this thesis will extract the information of sorts of hats,including the size and shape, in order to form the characteristic and classification with means of image enhancement, image thinning, feature extraction as well as algorithm of median filtering, edge detection and so on.. In addition, it also discusses system of hat image recognition with neural networks in order to achieve good effects. Eventually, according to the appropriate algorithm, comparison of image similarity will process with the help of image matching filtration to complete this project of system. The method of gray histogram transformation in hat recognition proves good effects to the results.The thesis firstly describes methods and procedure of hat image recognition in detail, and then it tests and analyzes the system. In the end, it summarizes and confirms the research orientation according to the results.